Influence of Heat Treatment on Defect Structures in Single-Crystalline Blade Roots Studied by X-ray Topography and Positron Annihilation Lifetime Spectroscopy
- 480 Downloads
Single-crystalline superalloy CMSX-4 is studied in the as-cast state and after heat treatment, with material being taken from turbine blade castings. The effect of the heat treatment on the defect structure of the root area near the selector/root connection is emphasized. Multiscale analysis is performed to correlate results obtained by X-ray topography and positron annihilation lifetime spectroscopy (PALS). Electron microscopy observations were also carried out to characterize the inhomogeneity in dendritic structure. The X-ray topography was used to compare defects of the misorientation nature, occurring in as-cast and treated states. The type and concentration of defects before and after heat treatment in different root areas were determined using the PALS method, which enables voids, mono-vacancies, and dislocations to be taken into account. In this way, differences in the concentration of defects caused by heat treatment are rationalized.
The single-crystalline, Ni-based superalloys are widely used for production of high-pressure and high-temperature turbine components in aerospace and energy industry sectors. Due to the extreme work conditions of blades, especially high mechanical and thermal stresses, the specific properties with low concentration of structural defects are needed.[1, 2, 3] Nowadays, the CMSX-4 single-crystalline superalloy is commonly used by industry; castings produced in this way are suitable candidates for defect characterization studies.
Directional dendritic solidification by the Bridgman technique is widely used for production of single-crystalline blades made from superalloys. Production technology and the complex shape of blade castings produce dendritic arrays that allow for the possibility of many defects to be produced during solidification. The defects may be related to the inhomogeneity of the chemical composition, morphology and size of γ′ particles, and the inhomogeneity of crystal orientation. Therefore, blades are subjected to complex heat treatment after casting, among others, to eliminate the chemical heterogeneity caused by directional dendritic crystallization. The heat treatment parameters are appropriately selected to create new γ/γ′ array (reprecipitation) with optimized properties, caused by more homogeneous morphology, size, and chemical composition. Some defects of macro-, micro-, and nano-scale are recreated in the treated blades, but some of them may be inherited from the as-cast state. All of them to some degree or other will influence the mechanical properties of blades and may cause damage during operation.[5, 6, 7, 8, 9] The macroscopic inhomogeneity in crystal orientation (e.g., subgrain boundaries or misoriented bands of dendrites, bended during crystallization) may be one of the inherited defects. These defects are generally related to the crystallization of complex shape casts.[4,10, 11, 12]
The inhomogeneity of crystal orientation may be characterized by X-ray diffraction topography in reflective geometry.[13,14] This imaging technique—which represents two-dimensional mapping of diffracted X-rays local intensity—is capable of providing information on the distribution and nature of structural defects in single-crystalline materials. The technique is sensitive to changes in local orientations of diffraction (atomic) planes and their spacing.[15,16] The method allows the visualization of the differences in crystal orientation (misorientation) of neighboring areas accurate to arc minutes. However, the analyzed areas must have a size of 1 mm2 or higher. The boundaries between such areas (macroscopic low-angle boundaries, LABs) are created by microscopic defects like dislocations, and additionally, they may be related to the vacancies. This means that the general description of defect structure of blades must refer to defects over a wide spatial size ranging from nanometer up to millimeter scale, which are mutually related to each other. Determining the relationship of different-scale defects, which is the basis of the multiscale analysis, helps to determine the reasons for their creation.
Additionally, the macroscopic inhomogeneity of crystal orientation may be related to sub-micrometer-scale defects, which are difficult or impossible to reveal by SEM or TEM methods due to their small analyzed area. The positron annihilation lifetime spectroscopy (PALS), using a macroscopic area, may enable it. PALS is a promising nondestructive method of quality control of technologically important materials employed in various fields of science and technology. Positrons interact for example with foreign atoms, points, or linear defects of structure. Positrons are trapped preferentially in atomic defects leading to an extended positron lifetime. The advantage of this method is the detection of small concentrations of defects that could not be detected by other methods. Positron techniques have been used for studies of defect behavior in the Ni3Al system and polycrystalline Ni-based superalloys, but there are no results as yet for investigation of defects inhomogeneity in the single-crystalline turbine blades.
It is emphasized that the structural defects are mainly created in the so-called critical areas of the blade. The single-crystalline blades possess critical areas with regard to the crystallization process (selector-root and root-airfoil connection areas, thin-walled areas) and to the operation loads (areas near the root-airfoil connection, thin-walled areas of trailing edge, tip area).[20,21] Regarding to the growth structure of dendrites set, the most affected area is situated in the root part of the blade, near the connection with the selector, where the step-change geometry of the castings occurs. The above justifies the selection of these regions for attention in this article. Its overarching aim of the study is to analyze and compare the defects structure of CMSX-4 single-crystalline blade root areas, located near the selector-root connection, in an as-cast state and after heat treatment using X-ray topography and the PALS method.
The heat treatment of part T was performed by several steps, consisting of convection heating to 950 °C (in a helium-protective atmosphere), radiation heating to 1350 °C (in vacuum), solution annealing, and finally aging. The temperature-time settings for annealing were: 1277 °C/4 h → 1287 °C/2 h → 1296 °C/3 h → 1304 °C/3 h → 1313 °C/2 h → 1316 °C/5 h → gas furnace quench, and for aging: 1140 °C/6 h (step 1) and 871 °C/20 h (step 2). The blade production and heat treatment process were performed in the Research and Development Laboratory for Aerospace Materials, Rzeszów University of Technology, Poland, using an industrial ALD Vacuum Technologies furnace.
The metallographic sections of SC and ST surfaces were studied by scanning electron microscopy (SEM), the X-ray diffraction topography, and Laue diffraction. A JEOL JSM-6480 microscope was applied for SEM observations using a backscattered electron (BSE) technique. A divergent beam of characteristic CuKα radiation, generated by a microfocus X-ray tube of the PANalytical system, was used for X-ray topography studies. The topograms were recorded on the AGFA Structurix D7 X-ray film in reflective geometry. The divergent beam of the characteristic radiation coming from a quasi-point (40 × 40 µm2) X-ray source illuminates the whole surface of the SC or ST surfaces. The sample (part C or T) coupled with the film oscillates about the axis located on the studied surface. The individual parts of the tested surface, meeting the Bragg condition, are successively recorded on the film. After over a dozen minutes of exposure, the diffraction image (topogram) of the whole analyzed surface was obtained on the film. If for areas close to surface SC of part C or ST for part T, the crystal lattices are rotated relative to each other, then their diffraction images will be mutually shifted in the topogram, creating different types of contrast. Similar displacement in the topograms was created by changes of d-spacing. Specific contrast is also created for other defects that occurred in the single-crystalline materials.[13,23] The misorientation angle may be calculated using the shift value in the topograms. The minimal misorientation angle, determined by X-ray topography method, is on the order of arc minutes, which are much lower than for the EBSD method.
Two samples in a sandwich arrangement (Figures 2(a) and 1(b)) for PALS measurements were prepared separately from each C-part and each T-part of the blades. The PALS measurements were performed in two areas (A1, A2 of Figure 2(b)) of each sandwich at room temperature with a conventional fast–fast spectrometer of 270-ps time resolution. The area A1 relates to the PSC area and A2 to the area outside it. The positron source with the activity of about 370 kBq, covered with a 5-μm Ni foil, was used. For each sandwich, a series of spectra was recorded. Then, the spectra were added together by means of a special procedure accounting for the drift of the zero-time channel. In this way, a resultant spectrum of very high statistics (at least 107 counts) was obtained. The data were analyzed using a least-square fitting procedure, which, contrary to the other existing programs, enables fitting not only a single spectrum but also a series of spectra. The simultaneous fitting leads to reduction of the number of the free fitting parameters because some of the parameters can have common values for all of the spectra of the series. In the calculations, all the measured spectra were fitted simultaneously with a model, directly implemented within the software code. All spectra were analyzed using the three-state trapping model describing the positron annihilations in the bulk material and two types of defects (Figure 2(b)). The three-state trapping model was described by three parameters, i.e., the positron lifetime in bulk material τb, the positron lifetime trapped by defects τ, and the positron trapping rate κ.
In Figure 2(b), the horizontal bars represent the free state of positrons, i.e., the delocalized state of each positron in bulk material and its localized states (dislocations and vacancy) in two types of defects. λb is the annihilation rate of the delocalized (free) positron from the bulk, λst and λt and λv are the annihilation rates from different type of defects, respectively. κv, κt, and κst are the trapping rates into these defects, which are proportional to the defect concentrations. ϑ are the trapping rate into dislocation-bound vacancy. The term δ is the detrapping (escape) rate.
All the measured spectra were fitted simultaneously at each parameter relating to the resolution or the contribution of source. This means that such a parameter was constrained to have the same value for each spectrum of the series of spectra analyzed together. The positron lifetimes relating to positron annihilation in the source were fixed. The “source” lifetimes were determined with help of a Si lifetime spectrum of very high statistics. The source contribution of 36.7 pct consisted of three components with lifetimes 125 ps, 386 ps, and 1.97 ns in the proportion 94.6:4.7:0.7, respectively.
In both topograms, the R-rings (RC, RT) of contrast with change in intensity of the rim of the PSC are visible. The RC ring (Figure 4(a)) consists of two sub-rings: the external one of SRC1 width with higher contrast intensity (Figure 4(e)) and the internal one of SRC2 width with lower contrast intensity (Figure 4(e)). The RT ring (Figure 4(b)) consists of two sub-rings: the external one of SRT1 width with lower contrast intensity (Figure 4(f)) and the internal one of SRT2 width with higher contrast intensity (Figure 4(f)). The contrast inversion can be observed during a hypothetical move from the center of the rings RC and RT toward the outside: In Figure 4(a), the bright contrast ring is visible first and the dark one is visible first in Figure 4(b).
There is also a visible dark band L1C in the topogram of the SC surface of the C-type part (Figures 4(a), (c), and (e)). Bright band L2T (Figures 4(b), (d), and (f)) may be observed in topograms obtained for T-type parts. This band appears after heat treatment. Fine parallel contrast bands (fine arrows, Figures 4(a), (c), and (e)), arranged along the m direction, appear in the topograms of the C-type part. However, in the topograms of T-type parts, the bands mentioned above are not visible. The m and n directions from Figure 4(a) are parallel to the secondary dendrite arms and to the  and  crystallographic directions. The blurred, wider bands H visible in the topograms from Figures 4(b), (d), and (f) are parallel to the h direction, which divides the angle between m and n in half. The G fragment (Figure 4(b)) with lack of contrast suggests that this area does not satisfy the Bragg’s condition and fragments of contrast are shifted creating an overlapped high contrast GT near the area G of the topogram.
The results obtained by the PALS method suggest that in the as-cast parts of the root (C-type), there is such a high concentration of defects that it is not possible to indicate differences in the defect structure between two selected areas A1 and A2. The determined values of positron lifetime (τ = 1.5 ns), obtained as a result of a numerical analysis of the experimental spectrum, allow us to identify the type of dominant defects. The determined lifetime corresponds to volumetric defects of the micro-voids type also called the free volumes.
Two types of defects were found in the T-type parts in both studied areas. Based on the positron lifetime calculations, the occurrences of mono-vacancies and dislocations were established. The defects concentration in the T-type parts (A1 and A2 areas) indicates that the dominant type of defects in both areas is vacancies. However, it should be noted that the concentration of each defect type in the discussed areas is different. It was observed that the vacancies concentration in the A1 area is about six times higher in relation to the dislocation concentration. The dislocation concentration in the A2 area is higher than in the A1 area, and additionally, the dislocation concentration in the A2 area is twice lower than the determined concentration of vacancies. The positron lifetimes for the vacancy in the area A1 (τv = 212 ps) and in the area A2 (τv = 220 ps) were determined. Similar calculations of positron lifetime were performed for dislocations in the area A1 (τv = 380 ps) and in the area A2 (τv = 410 ps).
The difference in the dendrites’ arrangement and arms’ length inside and outside the PSC area (Figure 3) is related to the difference in growth kinetics. The growth of dendrites is not disturbed inside the PSC area on the Pa plane (Figure 1(b)). Outside the PSC, the dendrite array in a whole root is conditioned by the rapid growth of the secondary dendrite arms on the Pa plane (Figure 1(b)).[4,20] The dendritic structure disorder in the K1 and K2 areas may be caused by local changes in the heat dissipation and by curvature of the solidification front in the narrow corner areas. It may also be the reason for the formation of contrast bands in topograms, separating the K1 and K2 areas (Figures 4(e), (c), and (a)), and this suggests a somewhat different crystal orientation of these areas in relation to other ones. Additionally, the secondary dendrite arms, growing in a transverse plane from the selector extension, do not reach into the K1 and K2 area, so the dendritic structure in these areas is different (Figures 3(a) and (b)) and often affected by the deflection process described in Reference 28. Structure alteration in the RC-ring region (Figure 3) similar to the specific shape of contrast of the PSC circle in topograms (Figures 4(a), (c), and (e)) is related to the circular shape of a selector continuer–root connection. This suggests that changes in the crystal orientation occur in this circle. The microstructure alteration visible on the macro-SEM images (e.g., K1, K2 areas, RC ring) and misoriented areas separated by boundaries similar to the LAB, called the “LAB-like” (L1, L2, R-type rings) visible in topograms (Figure 4), suggest crystal misorientation. As a result of heat treatment, some defects disappear, e.g., the LAB-like, visualized in topogram as L3C (Figures 4(a), (c), and (e)) but some remain (L1C − L1T, Figures 4(a) and (b) and 4(c) and (d)). A new defects formation after heat treatment, e.g., the LAB-like, marked as L2T (Figures 4(b), (d), and (f)), was also observed.
For the treated parts of the roots, the positron lifetime for vacancy in the area A1 (τv = 212 ps) and in the area A2 (τv = 220 ps) was determined. Similar calculations of positron lifetime were performed for dislocations in the area A1 (τv = 380 ps) and in the area A2 (τv = 410 ps). Differences in the given positron lifetime values can be related to different chemical environments of the defects. The differences in the concentration of defects estimated by PALS in the area A1 and A2 of T-type parts are probably related to different kinetics of the crystallization process in these areas. Electron microscopy studies of a dendritic structure (Figure 3) indicate the differences in the dendritic array of the A1 area (included in the PSC) and the A2 area. It may be observed in Figure 3(b). Probably heat treatment implies changes in the structure of point and linear defects; however, areas with a relatively homogeneous distribution of dendrites related to undisturbed crystallization of the PSC area (contained the A1 area) are characterized by a different concentration of point and linear defects, related to the area (outside the PSC) where the crystallization process was disturbed for a moment by the step-change geometry of the mold (near the Pa plane—Figure 1). The differentiated density of point defects in the studied areas may also be related to the inhomogeneity of the chemical composition in the defects environment, which suggests a different positron lifetime of the same-type defects. Thus, the treatment process does not homogenize the defect structure in the sub-microscopic scale. This requires further studies on a mechanism of γ′ re-precipitating and its atomic ordering during heat treatment.
Elimination of inhomogeneity, created during dendritic crystallization of a complex-shape cast, is one of the aims of heat treatment. A transition of the crystallization front through the areas of the step-change geometry causes momentary changes in the crystallization kinetics due to the rapid growth of the secondary dendrite arms, perpendicular to the withdrawal direction. This growth takes place in a thin layer of the root (near Pa—Figure 1) and occurs at a much higher growth rate than the withdrawal rate. As a result, a local disorder in crystal orientation and bending of dendrites, as well as the LAB-like formation, occur. The angle of misorientation of these boundaries can be estimated basing on the width of areas with increased or decreased contrast in the topogram. The areas of topograms of the increased contrast, derived from the SC surface, should be compared with the areas of the decreased contrast, derived from the ST surface, due to a contrast inversion resulting from the mirror-reflected SC and ST surfaces. In the case of the LAB L1C (Figures 4(a), (c), and (e)), the width SL1c is significantly smaller than the width SL1T (Figures 4(b), (d), and (f)), which means that the angle of misorientation is significantly smaller. Similarly, as in a previous case, the width of the SRC2 sub-ring is significantly smaller than the width of the SRT2 sub-ring (Figures 4(e) and (f)). It follows that heat treatment caused an increase of the misorientation angle at the RC and L1C boundaries. These macroscopic heterogeneities involve a defect generation at the micro- and nano-metric scale. Perfect heat treatment should, among others, eliminate dendritic segregation. Dissolution of the γ′ phase after heat treatment and then its re-precipitation eliminates some defects. The basic physical process controlling homogenization is diffusion. Macroscopic inhomogeneity in the crystal orientation of the millimeters- or tenths of millimeters-sized areas cannot be eliminated by diffusion due to limited processing time. In addition, changes in the crystal orientation do not directly correlate with the concentration gradient of the components. Therefore, the defects related to the crystal misorientation may remain after treatment.
A high concentration of volumetric defects revealed by PALS method in the as-cast parts of the roots may be related to a porosity. Nucleation of pores depending on the local stress level in the interdendritic melt may be driven by stress relaxation after pore nucleation that appeared near areas of step-changes in the mold geometry. The signal of positron annihilation from volumetric defects is high and prevents detecting vacancies and dislocations in an as-cast state. After heat treatment, the volumetric defects disappear and the vacancies with the dislocations may be detected by PALS.
Heat treatment does not eliminate most of the macroscopic defects that are associated with misorientation of dendrites. The misorientation angle of boundaries similar to low-angle boundaries (like-LAB) increases after heat treatment.
In the region near the rim of the selector continuer projection (PSC), a specific macroscopic distribution of crystal misorientation is present in both as-cast and heat-treated roots. This distribution has the character of a flexure mirror-like effect and does not disappear after heat treatment.
The as-cast regions of the blade roots, located near the selector continuer, contain high concentrations of micro-voids and pores; these make it impossible to determine by PALS the differences in vacancies and dislocations concentration.
It has been found by PALS that for heat-treated parts of the roots, there are differences in the concentration and type of defects between the area of PSC and beyond it. Inside it the concentration of dislocations is found to be lower than outside it. The concentration of vacancies inside the PSC area is higher.
Additionally, it was found that both inside and outside the PSC area after treatment, the concentration of vacancies is higher than the concentration of dislocations. This may be caused by migration of dislocations during treatment to the like-LAB and their annihilation. This assumption is consistent with the fact that the LAB-like angle of misorientation increases after heat treatment.
In this article, it has been demonstrated that the combination of X-ray topography and positron annihilation spectroscopy allows for a multiscale structural examination for the determination of defect structure, including the low-angle boundary (LAB) areas and nano- and micro-scale defects.
- 2.M.J. Donachie and S.J. Donachie: Superalloys - A technical Guide, 2nd ed., ASM International, Materials Park, 2002.Google Scholar
- 5.J.R. Li, J.Q. Zhao, S.Z. Liu, and M. Han: Superalloys 2008, TMS, 2008, pp. 443–51.Google Scholar
- 6.N. Stanford, A. Djakovic, B. Shollock, M. McLean, N. D’Souza, and P. Jennings: Superalloys 2004, TMS, 2004, pp. 719–26.Google Scholar
- 25.J.E. Miller, M. Strangwood, S. Steinbach, and N. Warnken: Proceedings of the 5th Decennial International Conference on Solidification Processing, Old Windsor, July 2017.Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.